[0001] The invention relates to a wavelength modulation spectroscopy method. It further
relates to a wavelength modulation spectroscopy system.
[0002] In wavelength modulation spectroscopy (WMS) the wavelength of the light of a tunable
light source, such as a diode laser, is modulated with a frequency f
0, while the wavelength is swept over a molecular absorption line of a gas component
of interest in a gas sample. As the light propagates along a measurement path through
the gas sample, wavelength dependent absorption converts some of the wavelength modulation
into an amplitude modulation of the light. Thus, the light will have an overtone spectrum
generated by the absorption, the harmonic content of the spectrum being dependent
on the width and shape of the molecular absorption line in the gas and the etalons
in the optical path of the measuring system. When the light then impinges onto a measuring
detector, for example a photodiode, the detector output contains AC components at
the modulation frequency f
0 and its higher harmonics Mf
0 (M = 2, 3, 4, etc.). Demodulating the detector output at one of said higher harmonics,
preferably at 2f
0, shifts the measurement from frequencies near DC, where the light source is noisy,
into a higher frequency range, where the noise is lower, thus improving the measurement
sensitivity.
[0003] In order to measure absolute gas concentrations, a suitable normalization method
is needed for compensating for general fluctuations in the emitted light intensity
and non-gas related transmission in the optical path of the measuring system. For
example, in in-situ measurements of trace gases in combustion environments where varying
dust loads, high temperature, gas turbulences etc. modulate the light in the kHz range,
it is important that the normalization is not distorted by the rapidly changing transmission
and turbulences in the measurement path.
[0004] Light which propagates through weakly absorbing gases is attenuated exponentially
according to the Beer-Lambert law:

where I is the intensity of the light after passing through the measurement path,
I
L is the intensity of the light emitted from the light source, T is a transmission
factor over the measurement path, which transmission factor stands for the wavelength
independent transmission including optical losses, α
i is the absorption coefficient of a gaseous species i with the concentration c
i, and L is the length of the measurement path. The absorption coefficient α
i is dependent on the light frequency ν (or the wavelength). For small optical absorption,
Equation 1 reduces to:

[0005] As mentioned above, wavelength modulation spectroscopy utilizes a rapid modulation
of the emitted light with a frequency f
0, while the wavelength is swept over a molecular absorption line of a gas component
of interest in the gas sample. The light impinging on the detector can then be written
as:

[0007] As can be seen from Equations 4 and 5, independent measuring of the non-gas related
transmission I
LT is needed to measure absolute gas concentrations.
[0008] The most straight forward method to measure the non-gas related transmission I
LT is to use a direct detection. The wavelength of the light is swept by a triangular
or sawtooth waveform over the absorption line of the gas component to be measured
wherein the beginning and the end of the scan are well separated from the absorption
peak. The measuring detector output is compared with the signal from a monitor detector
which directly monitors the output intensity of the light source. The direct detection
channel then detects the large triangular scan as a measure of the transmitted optical
power. The scan also includes a period where the light source is turned off in order
to provide an accurate zero irradiance reference. [
Applied Optics, Vol. 38, No. 36, pp. 7342-7354 (December 1999) and
Applied Optics, Vol. 44, No. 1, pp. 91-102 (January 2005)].
[0009] In wavelength modulation spectroscopy a combination of wavelength modulation and
direct detection can be used [
Applied Optics, Vol. 38, No. 21, pp. 4609-4622 (July 1999)]. This technique is mostly developed for atmospheric monitoring; to be used in harsh
industrial environment, the modulation rate has to be increased in order to place
the signal energy above that of the turbulent measuring medium.
[0010] In wavelength modulation spectroscopy an indirect measure of the non-gas related
optical transmission can be obtained by the use of the wavelength modulation signal
f
0, which makes it necessary to introduce a separate detection channel for the fundamental
frequency [
US 5,173,749]. An intentionally injected pilot tone at a higher harmonic Mf
0 of said wavelength modulation signal [
US 7,116,422] avoids the use of such a separate electronic channel. A drawback of this method,
however, is that the received pilot tone amplitude gives only information about the
transmission factor T rather than the detected non-gas related light intensity I
LT, thus I
L has to be measured separately, e. g. by division with a reference cell signal [
US 5,173,749], which introduces the necessity of an additional optical channel. Therefore, in
order to obtain I
LT directly, the modulation of the light source should also include turning off the
emitted light entirely.
[0011] Therefore, the invention seeks to provide a wavelength modulation spectroscopy method
and system, which effectively compensate variations in the emitted light intensity
and in the non-gas related transmission of the measurement path.
[0012] According to the invention this is achieved by the method defined in claim 1 and
the system defined in claim 7.
[0013] Preferred embodiments of the method and the system according to the invention are
specified in the remaining claims.
[0014] According to the present invention normalization is based on a burst signal, the
frequency of which lies above that of the turbulences and flame spectra in the measurement
path. The burst signal waveform is optimized to maximize the intensity modulation
effect while its amplitude is chosen to allow periodic interruption of the laser emission.
Moreover, the burst frequency is chosen so that a suitable overtone can be detected
by the same signal chain or channel as that of the analytical signal portion thereby
using preferably a down sampling scheme.
[0015] The present invention will be now described by way of example and with reference
to the accompanying drawings, in which:
- Figure 1
- shows an example for the light modulation with a burst and a sinusoidal signal;
- Figure 2
- shows the Fourier spectra of the detected burst and analytical signal portions;
- Figure 3
- illustrates the down sampling procedure; and
- Figure 4
- is a schematic block diagram of the system in accordance with the invention.
[0016] Figure 1 shows an example of the modulation of the intensity I
L of the light emitted from a light source, preferably a diode laser. The modulation
periodically alternates between a burst 1, 1' with a burst frequency f
n and a triangular or sawtooth sweep function 2 with a superimposed sinusoidal modulation
3 at a modulation frequency f
0. The purpose of the sweep function 2 is to allow light wavelength scan across an
absorption line of a gas component of interest. It is advantageous, although not necessary,
to apply the burst 1, 1' with different amplitudes before and after the sweep function
2 to allow measure of the eventual changes in the optical power vs. injection current
characteristics of the diode laser.
[0017] The waveform and the frequency f
n of the burst 1, 1' are preferably chosen to allow a settling time of the diode laser.
The optimum choice is a square wave. Furthermore, the burst frequency f
n is arranged so that an N-th harmonic Nf
n of the waveform coincides with an M-th harmonic Mf
0 of the modulation frequency f
0, at which M-th harmonic Mf
0 the analytical light portion I
AS (cf. Equation 5) is detected. Thus, for the purpose of normalization, a suitable
overtone can be detected by the same signal chain or channel as that for evaluation
the analytical signal portion. The choice of harmonics M and N is also made to allow
for an optimum relation in amplitude between the normalization signal and the analytical
signal, thus allowing an optimum dynamic range of the single signal chain.
[0018] The M-th harmonic of the detected analytical light portion (cf. Equation 5) can be
written as:

[0019] The detected light intensity as a result of the square wave burst modulation shown
in Figure 1 can be then written as:

where K depends on the bandwidths of the laser driver and the detector.
[0020] As the burst frequency f
n is arranged so that an N-th harmonic Nf
n, of the burst frequency f
n coincides with an M-th harmonic Mf
0 of the modulation frequency f
0, the following relation is given:

where N = 1, 3, 5,...
[0021] If another waveform configuration is used when the modulation and burst signals simultaneously,
a further aspect when determining the relation between these two frequencies f
0 and f
n is to avoid distortion due to overlap between frequency components of the measurement
and burst signal. In this case the bandwidth B of the measured signal at Mf
0 has to fulfill the following relation:

By inserting Equation 8 in Equations 6 and 7, respectively, one obtains:

and

[0022] Figure 2 shows the Fourier spectra of the detected burst and analytical signal portions
I
burst(f) and I
AS(f) for N = 7 and M = 2. The Nf
0 frequency components can be filtered and amplified before downsampling.
[0023] By performing downsampling at a sampling frequency F
s = (N+1)f
n, Equation 10 can be written as:

where n is a sample number. Similarly, Equation 11 becomes:

Since N/(N+1) > 1/2, aliasing takes place. Reconstruction of the discrete signals
given by Equations 12 and 13 gives:

[0024] Thus, by performing down sampling at a sampling frequency F
s = (N+1)f
n, the M-th harmonic of the detected analytical light portion I
AS,Nfn and the N-th harmonic I
burst,Nfn of the burst are both converted down to f
n due to the aliasing effect. This effect is shown in Figure 3.
[0025] Combination of Equations 12 and 13 yields the following formula for gas concentration:

As can be seen, the concentration c
i is no longer dependent on the non-gas related optical transmission I
LT.
[0026] The above method is especially advantageous to utilize an audio analog-to-digital
converter with a sampling frequency F
s = 192 kHz. This avoids the necessity of an extra downsampling stage. The burst frequency
is then f
n = 24 kHz, while the modulation frequency is f
0 = 84 kHz. The 7f
n burst and 2f
0 analytical signal fall both in a 168 kHz frequency band. Sampling at 192 kHz aliases
the 168 kHz band back to 24 kHz where they can be easily processed further.
[0027] Figure 4 shows a wavelength modulation spectroscopy system including a frequency
tunable light source 11 in form of a diode laser for generating light 12 in form of
a laser beam and of intensity I
L which is passed along a single optical path through a measuring volume 13 to a detector
14 for generating an output 15 indicative of the received light intensity I. The measuring
volume 13, which can be a sample cell or, in case of in-situ process measurements,
a gas-leading pipe, furnace, funnel or the like, contains a measuring gas (sample)
16, in which the concentration c
i of a specific gas component i is to be measured. The modulation of the diode laser
11 is switched by means of a switch 17 between the sweep signal 2 with the added modulation
signal 3 of the frequency f
0, provided by a waveform generator 18, and the burst signal 1, 1' turning on and off
the diode laser 11 at frequency f
n, provided by a burst generator 19. The frequencies f
0 and f
n are related such that Mf
0 = Nf
n, where N is an integer corresponding to a suitable harmonic of the burst signal 1,
1' and M is the harmonic of the modulation frequency f
0 where detection of the absorption in the measuring volume 13 will be made. The generated
laser light 12 is passed through the measuring volume 13 and picked up by the detector
diode 14. The output 15 of the detector 14 is filtered through a band-pass filter
20 with a centre frequency Mf
0 and then converted to digital format in an analog-to-digital converter 21 running
at a sampling frequency F
s = (N+1)f
n hence causing both the M-th harmonic of f
0 and the N-th harmonic of f
n to be folded or aliased back to frequency f
n. The down-converted detector output is then processed by the digital signal processing
unit 22 for calculating the concentration c
i of the specific gas component i to be measured. Due to a synchronisation unit 21
the signal processing unit 22 can separate the parts of the detector output related
to modulation generated by the burst generator 19 at the burst frequency f
n from those parts related to modulation by the waveform generator 18 at the modulation
frequency f
0.
1. A wavelength modulation spectroscopy method comprising the steps of:
- periodically sweeping the wavelength of a light source (11) over an interaction
feature of a sample (16) according to a sweep function (2),
- modulating the wavelength of said light source (11) with a modulation signal (3)
at a modulation frequency (f0), while the wavelength is swept over the interaction feature,
- further periodically modulating the intensity of said light source (11) at a wavelength
outside the interaction feature with a burst signal (1, 1'), where an N-th harmonic
(Nfn) of the burst frequency (fn) coincides with an M-th harmonic (Mf0) of said modulation frequency (f0),
- passing the light (12) of the light source (11) to the sample (16) for interacting
and thereafter to a detector (14),
- demodulating the detector output (15) at said M-th harmonic (Mf0) of said modulation frequency (f0), and
- normalizing the demodulated detector output (15) by calculating the ratio between
a demodulated detector output portion derived from the light (12) modulated with the
modulation signal (3) and another demodulated detector output portion derived from
the light (12) modulated with the burst signal (1, 1').
2. The method of claim 1 wherein before the step of normalizing, the demodulated detector
output (15) is down sampled at a sampling frequency (Fs) equal to the (N+1)-fold of the burst frequency (fn).
3. The method of claims 1 or 2 wherein N = 7 and M = 2.
4. The method of one of the preceding claims wherein the modulation signal (3) comprises
a sinusoidal.
5. The method of one of the preceding claims wherein the burst signal (1, 1') comprises
a square wave.
6. The method of one of the preceding claims wherein the sweep function (2) comprises
a sawtooth and wherein the amplitude of the burst signal (1, 1') is different before
and after the sweep (2).
7. A wavelength modulation spectroscopy system comprising:
- a wavelength tunable light source (11),
- first modulation means (18) for periodically sweeping the wavelength of the light
source (11) over an interaction feature of a sample according to a sweep function
(2) and modulating the wavelength of said light source (11) with a modulation signal
(3) at a modulation frequency (f0), while the wavelength is swept over the interaction feature,
- second modulation means (19) for periodically modulating the intensity of said light
source (11) at a wavelength outside the interaction feature with a burst signal (1,
1'), where an N-th harmonic (Nfn) of the burst frequency (fn) coincides with an M-th harmonic (Mf0) of said modulation frequency (f0),
- a detector (14) for detecting the light (12) of the light source (11) after interaction
with a sample (16) and producing a detector output (15),
- demodulation means (20) for processing the detector output (15) at said M-th harmonic
(Mf0) of said modulation frequency (f0), and
- evaluation means (22) for normalizing the demodulated detector output (15) configured
to calculate the ratio between a demodulated detector output portion derived from
the light (12) modulated with the modulation signal (3) and another demodulated detector
output portion derived from the light (12) modulated with the burst signal (1, 1').
8. The system of claim 7, further comprising between the detector (14) and the evaluation
means (22) an analog-to-digital converter (21) configured to downsample the demodulated
detector output (15) at a sampling frequency (Fs) equal to the (N+1)-fold of the burst frequency (fn).
9. The system of claims 7 or 8 wherein N = 7 and M = 2.
10. The system of one of claims 7 to 9 wherein the modulation signal (3) comprises a sinusoidal.
11. The system of one of claims 7 to 9 wherein the burst signal (1, 1') comprises a square
wave.
12. The system of one of claims 7 to 9 wherein the sweep function (2) comprises a sawtooth
and wherein the amplitude of the burst signal (1, 1') is different before and after
the sweep (2).